Skip to content
A set of utilities for checking and grading matplotlib plots
Python Jupyter Notebook
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
plotchecker Bump version Feb 17, 2018
.coveragerc Add methods for titles and axis labels Oct 8, 2015
.gitignore Add basic documentation Oct 8, 2015
.travis.yml Still trying to fix travis installation issue on python 3.4 Feb 17, 2018
Examples.ipynb Fix failing tests due to bit rot Feb 17, 2018
LICENSE Initial commit Jul 8, 2015 Add installation instructions Oct 9, 2015
flit.ini Enforce a recent matplotlib version Feb 17, 2018


Build Status Documentation Status

A set of utilities for checking and grading matplotlib plots. Please note that plotchecker is only compatible with Python 3, and not legacy Python 2. Documentation is available on Read The Docs.


To install plotchecker:

pip3 install plotchecker


The inspiration for this library comes from including plotting exercises in programming assignments. Often, there are multiple possible ways to solve a problem; for example, if students are asked to create a "scatter plot", the following are all valid methods of doing so:

# Method 1
plt.plot(x, y, 'o')

# Method 2
plt.scatter(x, y)

# Method 3
for i in range(len(x)):
    plt.plot(x[i], y[i], 'o')

# Method 4
for i in range(len(x)):
    plt.scatter(x[i], y[i])

Unfortunately, each of the above approaches also creates a different underlying representation of the data in matplotlib. Method 1 creates a single Line object; Method 2 creates a single Collection; Method 3 creates n Line objects, where n is the number of points; and Method 4 creates n Collection objects. Testing for all of these different edge cases is a huge burden on instructors.

While some of the above options are certainly better than others in terms of simplicity and performance, it doesn't seem quite fair to ask students to create their plots in a very specific way when all we've asked them for is a scatter plot. If they look pretty much identical visually, why isn't it a valid approach?

Enter plotchecker, which aims to abstract away from these differences and expose a simple interface for instructors to check students' plots. All that is necessary is access to the Axes object, and then you can write a common set of tests for plots independent of how they were created.

from plotchecker import ScatterPlotChecker

axis = plt.gca()
pc = ScatterPlotChecker(axis)

Please see the Examples.ipynb notebook for futher examples on how plotchecker can be used.

Caveats: there are many ways that plots can be created in matplotlib. plotchecker almost certainly misses some of the edge cases. If you find any, please submit a bug report (or even better, a PR!).

You can’t perform that action at this time.